🦞 Clawdian

The official OpenClaw × Obsidian plugin — bringing OpenClaw's full AI capabilities into your vault.

Chat with AI, edit notes inline, attach files as context, and run vault operations, all without leaving Obsidian. Powered by OpenClaw's agent system with tool use, memory, and custom personas.

Also works with any OpenAI-compatible API — OpenRouter, LM Studio, Ollama, or any self-hosted server.

Features

• Streaming chat with thinking/reasoning display
• Multi-conversation tabs with auto-save & AI-generated titles
• @ file mentions — attach any vault file as context
• Inline edit — edit text directly in the editor with AI (selection or cursor mode)
• Word-level diff — see exactly what AI changed before accepting
• Slash commands — /rewrite /translate /summarize /expand /fix /compact
• Custom commands — define your own prompt templates, access via / or right-click menu
• Usage-frequency sorting — most-used commands float to the top of the slash menu
• Current note context — AI always sees what you're working on
• Editor selection detection — select text, then ask about it
• Image paste & drag-drop — send screenshots to AI
• Input history — press ↑ in the input box to recall previous messages
• Model switcher — configure and switch between models
• Secure token storage — OS keychain with plaintext fallback
• Large file search — optional vault_search.py for 30K+ char files

Quick Start

1. Install

Download main.js, styles.css, and manifest.json from the latest release.

<your-vault>/.obsidian/plugins/clawdian/
├── main.js
├── styles.css
└── manifest.json

Enable in Obsidian → Settings → Community plugins.

2. Connect your API

Open Settings → Clawdian:

1. Gateway URL — your API endpoint
2. Token — your API key
3. Default model — the model name to use (e.g. gpt-4o, claude-sonnet-4, qwen-plus)
4. Click Test Connection

3. Add models to the switcher (optional)

In Settings → Clawdian → Custom models, add one model per line:

gpt-4o|GPT-4o
claude-sonnet-4|Sonnet 4
qwen-plus|Qwen Plus

Format: model-id|Display Name. These appear in the model dropdown in the chat toolbar.

4. Start chatting

Click the 🦞 icon in the sidebar. Done!

Provider Setup Examples

OpenRouter

Setting	Value
Gateway URL	https://openrouter.ai/api
Token	Your OpenRouter API key
Default model	anthropic/claude-sonnet-4

LM Studio

Setting	Value
Gateway URL	http://127.0.0.1:1234
Token	lm-studio (or leave empty)
Default model	Your loaded model name

Ollama

Setting	Value
Gateway URL	http://127.0.0.1:11434
Token	(leave empty)
Default model	llama3

OpenClaw Gateway (recommended)

OpenClaw provides the full experience: tool use, vault file operations, persistent memory, and agent personas.

Setting	Value
Gateway URL	http://127.0.0.1:18789
Token	Your OpenClaw token
Default model	openclaw/obsidian
Scopes header	operator.admin,operator.read,operator.write

Optional: Create a dedicated Obsidian agent for best results:

openclaw agents add obsidian --workspace "<your-vault-path>" --model "your-model"

Then place agent configuration files in your vault root. See examples/openclaw/ for templates:

• AGENTS.md — agent behavior rules
• SOUL.md — personality definition
• USER.md — your preferences
• TOOLS.md — tool usage guide

Usage

Chat

Action	Key
Send message	Enter
New line	Shift + Enter
Mention a file	Type @
Slash command	Type /
Stop generation	Click ■ button

Hover over any message to see action buttons (edit, regenerate, copy).

Inline Edit

1. Select text in the editor (or place cursor for insert mode)
2. Run command: Inline Edit (selection) or Inline Edit at cursor
3. Type your instruction (e.g. "translate to English", "make it concise")
4. Review the word-level diff → Enter to accept, Esc to reject

Slash Commands

Command	Description
/rewrite	Rewrite the current note
/translate	Translate selected text
/summarize	Summarize the current note
/expand	Expand selected content
/fix	Fix grammar and spelling
/compact	Compress conversation history to save tokens
/your-command	Any custom command you define in settings

Commands (Ctrl+P)

Command	Description
Open Clawdian	Open the chat panel
New Chat	Start a new conversation
Send selection to Clawdian	Send selected text to chat
Summarize current note	Generate a summary
Ask Clawdian about selection	Ask about selected text
Inline Edit (selection)	AI-edit selected text
Inline Edit at cursor	AI-insert at cursor

Custom Commands

Define your own prompt templates in Settings → Clawdian → Custom Commands:

1. Click + Add command
2. Fill in a name (e.g. polish) and a prompt (e.g. Polish the following text: {{text}})
3. Toggle In slash menu to access it via /polish
4. Toggle In right-click menu to access it by selecting text and right-clicking

{{text}} is replaced with your current selection when the command is triggered.

Optional: Large File Search

For vaults with large files (30K+ characters), Clawdian can use vault_search.py to search relevant passages instead of sending the entire file.

Setup

1. Copy tools/vault_search.py from this repo to anywhere on your machine
2. Install dependencies: pip install jieba
3. In Settings → Clawdian → Vault search script, enter the full path to the script

When enabled, large files attached via @ or as current note will be automatically searched for relevant content based on your question.

All Settings

Setting	Description	Default
Gateway URL	API endpoint URL	http://127.0.0.1:18789
Token	API authentication key	—
Default model	Model name for requests	openclaw/obsidian
Scopes header	OpenClaw-specific header (leave empty for other providers)	—
Custom models	Models for the dropdown switcher (id|label per line)	—
Include current note	Auto-attach focused note as context	✅
Show file actions	Display file action indicators in chat	✅
Vault search script	Path to vault_search.py (optional)	—
Conversations folder	Where to store chat history	Clawdian/conversations
Audit logging	Log all interactions to a file	❌

Data Storage

Data	Location
Conversations	<vault>/Clawdian/conversations/*.json
Markdown export	<vault>/Clawdian/conversations/md/*.md
Settings	<vault>/.obsidian/plugins/clawdian/data.json
Audit log	<vault>/Clawdian/audit-log.md (if enabled)

API Compatibility

Clawdian sends standard OpenAI-format requests to /v1/chat/completions. Any compatible API works:

Provider	Chat	Tool Use	Vault Operations
OpenClaw	✅	✅	✅
OpenRouter	✅	✅ (model-dependent)	❌
LM Studio	✅	❌	❌
Ollama	✅	❌	❌

Tool use and vault operations require a backend that supports function calling AND has file access to your vault. Without these, Clawdian works as a standard chat interface.

License

MIT — see LICENSE

Credits

Built by Zorba with Claude & Maya.

---

中文安装指南

第一步：下载插件文件

打开 最新 Release 页面，下载这三个文件：

• main.js
• styles.css
• manifest.json

也可以通过 BRAT 安装：添加仓库 Osamadhi/obsidian-clawdian，BRAT 会自动下载并支持后续更新。

第二步：放到 Obsidian 插件目录

在你的 Obsidian Vault 文件夹里，找到（或新建）这个路径：

你的Vault/.obsidian/plugins/clawdian/

把下载的三个文件放进去。

第三步：启用插件

打开 Obsidian → 设置 → 第三方插件 → 已安装插件 → 找到 Clawdian → 打开开关。

如果没看到，点一下「刷新」按钮。

第四步：填写连接信息

打开 Obsidian 设置 → Clawdian：

设置项	说明	示例值
Gateway URL	OpenClaw 服务地址	http://127.0.0.1:18789
Token	OpenClaw 访问令牌，由服务管理员提供	—
Default model	默认使用的模型	openclaw/obsidian

填完点 Test Connection，显示绿色 ✅ 即连接成功。

Token 和 Gateway URL 由 OpenClaw 服务管理员提供。如需了解 OpenClaw 部署，参见 OpenClaw 文档。

第五步：开始使用

左侧边栏点 🦞 龙虾图标，打开 Clawdian，开始对话。

常用操作：

• Enter 发送，Shift+Enter 换行
• 输入 @ + 文件名，将某篇笔记作为上下文发送给 AI
• 在编辑器中选中文字，切到 Clawdian 直接提问
• 输入 / 查看内置及自定义快捷命令

遇到问题请在 GitHub Issues 提交反馈。